Racial Differences in the Impact of HMO Coverage of
Diabetes Blood Glucose Monitors on Self-Monitoring
Connie A. Mah, M.S.Department of Ambulatory Care & Prevention
Harvard Medical School & Harvard Pilgrim Health [email protected]
AcknowledgementsCo-Authors:
Stephen B. Soumerai, ScD Alyce S. Adams, PhD Dennis Ross-Degnan, ScD
Funders: AHRQ Harvard Pilgrim Health Care Foundation
Consultants: Fang Zhang, PhD John D. Piette, PhD James Meigs, MD, MPH
BackgroundDisparities in diabetes health & health
care use Blacks have worse glycemic control Blacks at greater risk of adverse outcomes
Differences in diabetes quality of care & self-management Blacks receive lower quality of care Blacks face greater barriers to self-management
Coverage PolicyObjective
To reduce cost as a barrierTo increase access to high quality care
Mandated in over 38 states
Provision of glucose monitoring devices & supplies
Findings from parent evaluation study*Increased rates of SMBG
*Soumerai, Mah, Zhang, et al., Archives of Internal Medicine 2004 (164(6):645-52)
Key Research QuestionsDoes this policy have the potential to
narrow some of the racial disparity in diabetes self-management?
Does providing free home glucose monitors have differential impacts on self-monitoring for black and white diabetes patients?
PurposeTo determine whether policy increased
initial trials of self-monitoring among black versus white patients
To investigate whether patients who initiated self-monitoring after the policy continued to self-monitor thereafter
Research Design & SettingResearch Design
Longitudinal, retrospective cohort analysis
Study Setting Harvard Vanguard Medical Associates (HVMA)Harvard Pilgrim Health Care (HPHC)
HPHC Coverage PolicyStart of Implementation: Oct 1,1993
Objective To motivate diabetes patients to start monitoring their
blood glucose
Policy Benefits Provision of glucose monitoring devices Self-management training & education Lower copay for up to 3 months’ supply of test strips
(≤$5/script)
Study Cohort
Definition of Diabetes:≥ 1 hospital discharge Dx; or ≥ 2 outpatient Dx; or ≥ 1 insulin or oral sulfonylurea Rx
Black or White race only
Continuous enrollment (1992-1996)
N=2,275 adult patients
HVMA Data Sources (1992-1996)
Ambulatory Medical Records
Membership/ Enrollment
Pharmacy Claims
Clinical Encounters
HbA1c Lab Results
Race Identification
Drug Therapy
Dispensed Test Strips
Demographic Info Home Address (linked by census
block group)
Days Enrolled
PATIENT STUDY ID
Census File
Key MeasuresMain independent measure
RaceRace (black v. white)Outcome measures
Incidence of SMBGIncidence of SMBG (≥1 strip) Discontinuation of SMBGDiscontinuation of SMBG (>180 days w/out strips)
Covariates FixedFixed (age, sex, census-derived median HH income &
educational level, drug type, BMI, HbA1c test, primary health site)
Time-VaryingTime-Varying (mean HbA1c values in prior month, # MD visits per month)
Main Analytical MethodsKaplan-Meier & Log-Rank Tests
Cumulative rates of initiation of SMBGCumulative rates of discontinuation of SMBG
Extended Segmented Cox ModelsAdjusting for patient-level fixed & time-varying
covariatesRelative (hazard) rates of initiation of SMBG
(blacks relative to whites)
Pre-Policy Patient DifferencesAny
InsulinOral Sulfonylurea
Only
Black
(n=264)
White
(n=613)
Black
(n=250)
White
(n=739)
Female 63.3% 50.4% 57.6% 44.0%
Mean Age 52±12 52±14 53±11 60±11
Mean BMI 32.7±7.1 30.5±7.5 32.0±6.3 32.1±7.2
Mean HbA1c 9.5±1.7 9.0±1.4 9.1±1.9 8.1±1.4
% Any SMBG 64.8% 77.8% 34.0% 30.8%
Bold denote p<0.05
Any Pre-Policy Insulin Treatment (N=567)
0
0.2
0.4
0.6
0.8
1
Ap
r-92
Au
g-9
2
De
c-9
2
Ap
r-93
Au
g-9
3
De
c-9
3
Ap
r-94
Au
g-9
4
De
c-9
4
Ap
r-95
Any Pre-Policy Oral Treatment (N=904)
0
0.2
0.4
0.6
0.8
1
Ap
r-92
Au
g-9
2
De
c-9
2
Ap
r-93
Au
g-9
3
De
c-9
3
Ap
r-94
Au
g-9
4
De
c-9
4
Ap
r-95
Black White
% w
ith
SM
BG phase-inphase-in
Est. Haz Ratio 95% CI
Pre-Policy Difference
(Black vs. White) 1.09 0.84,1.40
Post-Policy Difference
(Black vs. White) 1.35* 1.05,1.72
Censoring at first insulin use or never initiated SMBG
* p <0.05
Controlling for age, glycemic control, time-dependent drug use, and time-dependent number of physician visits
0102030405060708090
100
6 Months 12 Months 18 Months
Months since SMBG Initiation
% S
MB
G D
isco
nti
nu
atio
n
Black (n=59)
White (n=134)
p<0.05
Post-Policy SMBG Initiators
Discontinuation = >180 days without test strip use
SummaryTrials of self-monitoring in post-policy
increase in SMBG
greater for blacks on oral therapy
Persistence after initiation of SMBG in post-policy
short-lived
LimitationsMissing race data
Important unmeasured factorsSocio-cultural factors (attitudes, perceptions,
cultural beliefs/values)Duration of illnessIntensity of medication use
Single HMO